Surface electromyography signal denoising via EEMD and improved wavelet thresholds

Area of Science:

• Biomedical Engineering
• Signal Processing
• Rehabilitation Technology

Background:

• Accurate prosthetic limb control relies on high-quality surface electromyography (sEMG) signals.
• sEMG signals are inherently nonlinear, nonstationary, and susceptible to noise interference.
• Existing denoising methods may not fully address the complexities of sEMG signal noise.

Purpose of the Study:

• To propose and evaluate a novel sEMG denoising method for enhanced prosthetic limb control.
• To address the challenges of noise in nonlinear and nonstationary sEMG signals.
• To improve the reliability and accuracy of prosthetic limb movement control.

Main Methods:

• A hybrid denoising approach integrating Ensemble Empirical Mode Decomposition (EEMD) and wavelet thresholding.
• EEMD decomposes the sEMG signal into intrinsic mode functions (IMFs).
• Wavelet transform methods are applied to extract useful components from noise-dominated IMFs and denoise other IMFs using improved thresholding techniques.

Main Results:

• The proposed EEMD-wavelet method effectively removes random noise from sEMG signals.
• Experimental validation across various muscles and motions demonstrated superior performance compared to conventional wavelet and EMD methods.
• The denoising algorithm successfully retained essential signal components for accurate control.

Conclusions:

• The developed hybrid denoising technique significantly enhances sEMG signal quality.
• This method offers a robust solution for real-time prosthetic limb control applications.
• Improved sEMG signal processing is crucial for advancing prosthetic functionality and user experience.